I don't think science as we currently understand it is capable of grasping consciousness in the full sense of that word, but IIT is the best out there. But it's still science as usual and a dead end insofar as it supposes neuronal activity of the brain is all there is to it. That's just mistaking the part for the whole. It's a rather bizarre presumption to make given IIT's starting point.
I want to push back against IIT being "science as usual". In the first place it's unusual in beginning from subjective experience to derive its "axioms". But even more unusually, as I learned from Mike's comments below, it turns out it's far more idealist than I had realised from the book (or perhaps it has moved in that direction since the book was written). Quoting the post Mike linked,
"The message is repeated over and over: my neurons do not truly exist. They exist in a derived, extrinsic sense—as stable appearances that other conscious observers, such as neurosurgeons, can use as handles of control over my experiences—but they do not intrinsically exist. What intrinsically exists is my conscious point of view."
Which is a pretty fantastic claim! IMO it's quite encouraging that serious scientists are daring to think so far beyond the mechanistic reductionist box.
I totally agree about it being unusual for its axioms, and definitely way ahead of the game as far as scientific theories go in conceding the experiential nature of consciousness. I've heard Koch saying things that sound idealistic recently, but I was under the impression that was just him speaking for himself. The theory is a different thing, I would imagine. But maybe I'm wrong? I haven't read the book, so you'd know better than I do!
On the other hand, I get the distinct sense that something fishy is going on from articles like this:
"IIT says you have a system—a bunch of wires or neurons—and it’s the extent to which they have causal power upon themselves. You’re always looking for the maximum causal power that the system can have on itself. That is ultimately what consciousness is. It’s something very concrete. If you give me a mathematical description of a system, I can compute it, it’s not some ethereal thing."
Then he says in the same article:
"I actually see you on the screen, there are lights in the image; inside my brain, I can assure you, there are no lights, it’s totally dark. My brain is just in a goo. So it’s not my brain that sees; it’s consciousness that sees. It’s not my brain that makes a decision, it’s my consciousness that makes a decision. They’re not the same."
Ooookay. I just get the sense there's a bait and switch going on here. The way they describe the theory isn't the same as the theory itself, which is fundamentally a measure of brain activity, a reduction of consciousness to mathematics. Here he's talking about IIT being objectively measurable:
"The experiments were trying to predict where the “neural footprints of consciousness,” crudely speaking, are. Are they in the back of the brain, as integrated information theory asserts, or in the front of the brain, as global neuronal workspace asserts? And the outcome was very clear—two of the three experiments were clearly against the prefrontal cortex and in favor of the neural footprint of conscious being in the back."
So consciousness is in the back of the brain. Sounds pretty standard to me. And confused.
From what I recall, the theory and the book do go on to some of the more radical philosophical implications about ontology, but I don't think it (explicitly) went as far as the kind of idealism I mentioned. Or possibly I missed it, since I'm often a little distracted when listening to audiobooks.
Is the reduction of consciousness to mathematics so absurd though? Mathematics is, after all, a basically Platonist discipline essentially concerned with immaterial realities/ideas. It's something I've been finding funny about physicalism lately: physicalism says everything is physical, but forgets that physics rests on mathematics, and mathematics is obviously anything but physical!
I think the distinction in IIT between consciousness and its neural correlates is kind of vaguely Aristotelian. They are the same but distinct. It's the form of the brain (or certain brain regions). IIRC, Koch actually says at one point that the notion of information in IIT is closer to Aristotle's notion of 'form', rather than Shannon information.
I know what you mean about audiobooks. I tend to listen to them on long road trips, otherwise I can't focus.
Platonism doesn't find mathematics the ultimate reality though. Only a stepping stone, an educational tool that leads to the ultimate qualitative reality.
I think when it comes to physicalism, everything is physical means there is nothing that isn't physical. In other words, it's a redefinition of the physical so that anything you can possibly name is just called physical. That's how information gets to be called physical. I think it's a deflationary metaphysics that works by definitional fiat.
I'm not sure about Aristotle. Wouldn't he have said the physical correlate is our entire body because the body is that which is under our agential control?
I thought Platonism saw mathematical objects as part of the real of forms? But it's been a while since I read Plato and you know him better than me.
For myself in any case, I think mathematical objects are a type of eternal object, and also that there's no clear divide between the qualitative and the quantitative. Partly because I think we can see qualities structurally, and partly because we experience mathematical objects qualitatively.
I agree about physicalism generally just defining everything as physical, but I think whenever you look closer at physicalist accounts, there seems to be an implicit platonist or Aristotelian element left that does all the heavy lifting. Like the functions in functionalism, implicitly brings in final and formal causes. The information in computationalism and IIT brings in formal causes again. The mathematics in science generally smuggles in a touch of (mathematical) platonism.
Well, iiirc Aristotle does at times consider that the soul might actually be a particular part of the body. I think he was undecided on its exact location and nature. Although perhaps he was just responding to the views of others. And I haven't read De Anima, so he might have set the record straight there.
In any case, it's probably a mistake to equate his idea of a soul with IIT's idea of a MICS/Whole. But I find the parallel helpful.
Yeah, mathematical objects are forms, but intermediary forms. If you're familiar with the divided line in the Republic, they're on the third segment from the bottom. And Plato would agree, mathematical objects are both qualitative and quantitative, although you have to have knowledge of the Good to fully understand how they're qualitative. They're something in-between pure forms and sense perception because we rely on diagrams and visual aids to do math, yet what we're looking at isn't what we're referring to (given that a line has no breadth and is invisible).
Yes, there's definitely an implicit "form" going on in functionalism. Functionalists tend not to like it when you point that out, though, since any theory of forms is seen as ludicrous.
For Aristotle I think he considered the whole body the form of the soul, but also he thought the heart was the most important. Not quite the seat of the soul in the sense of being identical to it, but the organ on which all the other organs depend.
I want to read two of your linked posts before I comment on panpsychism or experience. About IIT, I think it may be a necessary component but doubt it is sufficient.
Based on the one example we know is conscious — brains — I think a *physical* complex network is obviously necessary and further that synaptic connections (rather than mere interconnections) are key. (I once read a neurophysicist describe synapses as the most complex biological engine we know of.) I've long been skeptical that a software simulation of such a network would work, though LLMs do give me some pause. (If interested, I can point you to a number of posts I wrote here last year discussing why. They're in the "My Best Guess" newsletter starting last August.)
WRT group consciousness, indeed, and you might find this recent post about "bio-behavioural synchrony" interesting:
The post is more about one-on-one synchronization, but in my comment there I asked about sporting events and political rallies as examples of many-to-many and one-to-many synchronization. It's a fascinating topic, and for me, explains a lot of life experiences.
I look forward to hearing your thoughts on those posts :) I agree, IIT feels like a good contribution, but we're not there yet.
That's interesting that you think it's synapse specific. I'm open to the possibility, although I haven't about heard how complex they are. Although I did read an article about how individual neurons in the brain learn via predicting their own future states, which to me shows we're already dealing with an intelligent organism.
Could you link to those posts on LLMs? That's a topic I'm very interested in.
Thanks for sharing that post, that is fascinating.
I think the synapses are important because they're a big part of learning though we now know the brain does rewire itself as well.
To be clear, my skepticism and the posts I mentioned are about computationalism in general. It's that LLMs have seemed to offer contrary evidence, though more recently we seem to be seeing their limitations (thus preserving my skepticism 😁).
This is a good summation of Koch's pitch. I read his book some years ago, mostly because I felt like I needed to get the best sales pitch for IIT I could find. Can't say I was a fan before reading it, and the book itself didn't help.
My first issue with IIT is it takes introspective impressions as its foundation. I think we have enough data from psychological research to see that as a foundation of sand.
Another issue is it seems disconnected from cognitive neuroscience, ignoring the work that's been done on which regions are involved in introspection, attention, affective reactions, and many other things. And I find the postulates vague and redundant.
I agree about the exclusion postulate. It feels like the theory architects decided they needed to save appearances and ban talk of group consciousness and overlapping consciousnesses. And I agree that integration is crucial, but not sufficient.
Finally, any theory that predicts zombies (feed forward equivalents) and undetectable consciousness (such as Scott Aaronson's examples with trivially high phi) seems to have serious falsifiability issues. Some of the stipulations just seem arbitrary to meet a particular philosophical expectation.
When IIT first came out, it was often discussed as panpsychist in its orientation. A few years ago, Jonathan Birch noticed that the more recent papers had taken an idealist turn. Koch himself in a recent interview on Sean Carroll's blog largely confirmed this change. Overall, it just makes IIT feel more like speculative philosophy than a hard nosed scientific theory.
I disagree about it being problematic to take introspective impressions as a foundation. If we were wanting to simply investigate the workings of the brain, then you're right, introspection is a pretty unreliable source. But if we're wanting to look at the nature of subjective experience itself, then we have to start with introspection so that we thoroughly know what it is we're trying to explain. Even if it's misleading or illusory in one way or another, we need to understand that manifest image (please correct me if I'm using that wrong) in order to be able to explain why it appears that way.
That's interesting that it's taken an idealist turn. What does that look like? I've not really got my head around what idealism is supposed to mean (besides Berkeley's version).
I think it's on the border between science and philosophy, but that's to be expected of consciousness studies imo, since it's pre paradigm. And in general the border between them is less solid than is often supposed.
I agree that the manifest image has to be accounted for. But IIT takes it as a guide to the substrate that a consciousness has to exist in. For all the reasons you discuss (like the arbitrary requirement about granularity) this just seems like assumptions to get a desired result.
Definitely every scientific theory of consciousness is both a philosophy and an empirical model. But as Birch covers in his series, IIT's philosophy is well outside of the current Overton window. Should that window shift? I'd say yes, if there are good reasons for it. I'm not sure what those would be in this case.
"The message is repeated over and over: my neurons do not truly exist. They exist in a derived, extrinsic sense—as stable appearances that other conscious observers, such as neurosurgeons, can use as handles of control over my experiences—but they do not intrinsically exist. What intrinsically exists is my conscious point of view."
This reminds me of a bit in Aristotle's Metaphysics where he says that the parts of a substance (eg parts of a living body) do not have substantial existence themselves, but only potential existence, so that there can be no substance composed of substances (for my own future reference, it's in Zeta 16). I don't think Aristotle was saying anything as radical as that paper was (pretty sure he wasn't an idealist), but it's interesting to see the similarity.
I suppose if IIT involves all matter being part of a conscious whole, then the similarity on this point is very close. And I think Aristotle kind of had a point: If we're looking at things in terms of their behaviour as part of a whole, it won't do to ignore the full context of the whole whole. You can't really understand a heart apart from its context within an animal for example.
The difficulty is, how do you draw the line for where a whole ends? I suppose that's where the exclusion principle comes in again, and Aristotle has his own criteria.
But if we take a relational metaphysics instead of a substantial one, then things look quite different. We can look at things in their full contexts without needing to draw any line on where their substance/whole ends, while still recognising that certain regions are more relevant than others.
This kind of relates to what you said in your recent post about what it's like to be a bat, seeing consciousness as kind of being "hooked up" to perceive yourself. Rather than seeing self-causal loops/consciousness as fundamentally different from our perception of external reality, as IIT does, we can see our self-perception and perception of the world as the same kind of thing. Hope that makes sense?
I think the Overton window can shift in response to empirical validation. Of course, metaphysical aspects of IIT and other theories of consciousness cannot be empirically tested directly, but for those aspects that can, these should imo count as evidence for the metaphysical elements. At least if one metaphysical system fits more nicely with all the evidence. As an example, imo evolution has major implications for metaphysics and the existence or nature of God.
I definitely think our reflective perceptions aren't fundamentally different than our external ones, so I'm onboard. When I'm trying to steelman IIT, I actually take all the talk of "having causal power over itself" as basically an alternate way of describing self reflection. But I find the language so convoluted and ambiguous, I can't be sure that's what they mean.
I don't think science as we currently understand it is capable of grasping consciousness in the full sense of that word, but IIT is the best out there. But it's still science as usual and a dead end insofar as it supposes neuronal activity of the brain is all there is to it. That's just mistaking the part for the whole. It's a rather bizarre presumption to make given IIT's starting point.
I want to push back against IIT being "science as usual". In the first place it's unusual in beginning from subjective experience to derive its "axioms". But even more unusually, as I learned from Mike's comments below, it turns out it's far more idealist than I had realised from the book (or perhaps it has moved in that direction since the book was written). Quoting the post Mike linked,
"The message is repeated over and over: my neurons do not truly exist. They exist in a derived, extrinsic sense—as stable appearances that other conscious observers, such as neurosurgeons, can use as handles of control over my experiences—but they do not intrinsically exist. What intrinsically exists is my conscious point of view."
Which is a pretty fantastic claim! IMO it's quite encouraging that serious scientists are daring to think so far beyond the mechanistic reductionist box.
I totally agree about it being unusual for its axioms, and definitely way ahead of the game as far as scientific theories go in conceding the experiential nature of consciousness. I've heard Koch saying things that sound idealistic recently, but I was under the impression that was just him speaking for himself. The theory is a different thing, I would imagine. But maybe I'm wrong? I haven't read the book, so you'd know better than I do!
On the other hand, I get the distinct sense that something fishy is going on from articles like this:
https://nautil.us/what-counts-as-consciousness-579734/
First he says:
"IIT says you have a system—a bunch of wires or neurons—and it’s the extent to which they have causal power upon themselves. You’re always looking for the maximum causal power that the system can have on itself. That is ultimately what consciousness is. It’s something very concrete. If you give me a mathematical description of a system, I can compute it, it’s not some ethereal thing."
Then he says in the same article:
"I actually see you on the screen, there are lights in the image; inside my brain, I can assure you, there are no lights, it’s totally dark. My brain is just in a goo. So it’s not my brain that sees; it’s consciousness that sees. It’s not my brain that makes a decision, it’s my consciousness that makes a decision. They’re not the same."
Ooookay. I just get the sense there's a bait and switch going on here. The way they describe the theory isn't the same as the theory itself, which is fundamentally a measure of brain activity, a reduction of consciousness to mathematics. Here he's talking about IIT being objectively measurable:
"The experiments were trying to predict where the “neural footprints of consciousness,” crudely speaking, are. Are they in the back of the brain, as integrated information theory asserts, or in the front of the brain, as global neuronal workspace asserts? And the outcome was very clear—two of the three experiments were clearly against the prefrontal cortex and in favor of the neural footprint of conscious being in the back."
So consciousness is in the back of the brain. Sounds pretty standard to me. And confused.
From what I recall, the theory and the book do go on to some of the more radical philosophical implications about ontology, but I don't think it (explicitly) went as far as the kind of idealism I mentioned. Or possibly I missed it, since I'm often a little distracted when listening to audiobooks.
Is the reduction of consciousness to mathematics so absurd though? Mathematics is, after all, a basically Platonist discipline essentially concerned with immaterial realities/ideas. It's something I've been finding funny about physicalism lately: physicalism says everything is physical, but forgets that physics rests on mathematics, and mathematics is obviously anything but physical!
I think the distinction in IIT between consciousness and its neural correlates is kind of vaguely Aristotelian. They are the same but distinct. It's the form of the brain (or certain brain regions). IIRC, Koch actually says at one point that the notion of information in IIT is closer to Aristotle's notion of 'form', rather than Shannon information.
I know what you mean about audiobooks. I tend to listen to them on long road trips, otherwise I can't focus.
Platonism doesn't find mathematics the ultimate reality though. Only a stepping stone, an educational tool that leads to the ultimate qualitative reality.
I think when it comes to physicalism, everything is physical means there is nothing that isn't physical. In other words, it's a redefinition of the physical so that anything you can possibly name is just called physical. That's how information gets to be called physical. I think it's a deflationary metaphysics that works by definitional fiat.
I'm not sure about Aristotle. Wouldn't he have said the physical correlate is our entire body because the body is that which is under our agential control?
I thought Platonism saw mathematical objects as part of the real of forms? But it's been a while since I read Plato and you know him better than me.
For myself in any case, I think mathematical objects are a type of eternal object, and also that there's no clear divide between the qualitative and the quantitative. Partly because I think we can see qualities structurally, and partly because we experience mathematical objects qualitatively.
I agree about physicalism generally just defining everything as physical, but I think whenever you look closer at physicalist accounts, there seems to be an implicit platonist or Aristotelian element left that does all the heavy lifting. Like the functions in functionalism, implicitly brings in final and formal causes. The information in computationalism and IIT brings in formal causes again. The mathematics in science generally smuggles in a touch of (mathematical) platonism.
Well, iiirc Aristotle does at times consider that the soul might actually be a particular part of the body. I think he was undecided on its exact location and nature. Although perhaps he was just responding to the views of others. And I haven't read De Anima, so he might have set the record straight there.
In any case, it's probably a mistake to equate his idea of a soul with IIT's idea of a MICS/Whole. But I find the parallel helpful.
Yeah, mathematical objects are forms, but intermediary forms. If you're familiar with the divided line in the Republic, they're on the third segment from the bottom. And Plato would agree, mathematical objects are both qualitative and quantitative, although you have to have knowledge of the Good to fully understand how they're qualitative. They're something in-between pure forms and sense perception because we rely on diagrams and visual aids to do math, yet what we're looking at isn't what we're referring to (given that a line has no breadth and is invisible).
Yes, there's definitely an implicit "form" going on in functionalism. Functionalists tend not to like it when you point that out, though, since any theory of forms is seen as ludicrous.
For Aristotle I think he considered the whole body the form of the soul, but also he thought the heart was the most important. Not quite the seat of the soul in the sense of being identical to it, but the organ on which all the other organs depend.
I want to read two of your linked posts before I comment on panpsychism or experience. About IIT, I think it may be a necessary component but doubt it is sufficient.
Based on the one example we know is conscious — brains — I think a *physical* complex network is obviously necessary and further that synaptic connections (rather than mere interconnections) are key. (I once read a neurophysicist describe synapses as the most complex biological engine we know of.) I've long been skeptical that a software simulation of such a network would work, though LLMs do give me some pause. (If interested, I can point you to a number of posts I wrote here last year discussing why. They're in the "My Best Guess" newsletter starting last August.)
WRT group consciousness, indeed, and you might find this recent post about "bio-behavioural synchrony" interesting:
https://neuroscienceandpsy.substack.com/p/the-sandman-effect
The post is more about one-on-one synchronization, but in my comment there I asked about sporting events and political rallies as examples of many-to-many and one-to-many synchronization. It's a fascinating topic, and for me, explains a lot of life experiences.
I look forward to hearing your thoughts on those posts :) I agree, IIT feels like a good contribution, but we're not there yet.
That's interesting that you think it's synapse specific. I'm open to the possibility, although I haven't about heard how complex they are. Although I did read an article about how individual neurons in the brain learn via predicting their own future states, which to me shows we're already dealing with an intelligent organism.
Could you link to those posts on LLMs? That's a topic I'm very interested in.
Thanks for sharing that post, that is fascinating.
I think the synapses are important because they're a big part of learning though we now know the brain does rewire itself as well.
To be clear, my skepticism and the posts I mentioned are about computationalism in general. It's that LLMs have seemed to offer contrary evidence, though more recently we seem to be seeing their limitations (thus preserving my skepticism 😁).
That said:
https://logosconcarne.substack.com/p/brains-are-nothing-like-computers
https://logosconcarne.substack.com/p/brains-are-not-algorithmic
https://logosconcarne.substack.com/p/digital-emulation
https://logosconcarne.substack.com/p/digital-simulation
https://logosconcarne.substack.com/p/analog-vs-digital
https://logosconcarne.substack.com/p/computation-vs-evaluation
https://logosconcarne.substack.com/p/is-reality-a-computation
https://logosconcarne.substack.com/p/computation-vs-playback
Which distill and recap several dozens of posts from my old WordPress blog. If you're interested, this tag offers a (hopefully mostly complete) list:
https://logosconcarne.com/tag/consciousness/
This is a good summation of Koch's pitch. I read his book some years ago, mostly because I felt like I needed to get the best sales pitch for IIT I could find. Can't say I was a fan before reading it, and the book itself didn't help.
My first issue with IIT is it takes introspective impressions as its foundation. I think we have enough data from psychological research to see that as a foundation of sand.
Another issue is it seems disconnected from cognitive neuroscience, ignoring the work that's been done on which regions are involved in introspection, attention, affective reactions, and many other things. And I find the postulates vague and redundant.
I agree about the exclusion postulate. It feels like the theory architects decided they needed to save appearances and ban talk of group consciousness and overlapping consciousnesses. And I agree that integration is crucial, but not sufficient.
Finally, any theory that predicts zombies (feed forward equivalents) and undetectable consciousness (such as Scott Aaronson's examples with trivially high phi) seems to have serious falsifiability issues. Some of the stipulations just seem arbitrary to meet a particular philosophical expectation.
When IIT first came out, it was often discussed as panpsychist in its orientation. A few years ago, Jonathan Birch noticed that the more recent papers had taken an idealist turn. Koch himself in a recent interview on Sean Carroll's blog largely confirmed this change. Overall, it just makes IIT feel more like speculative philosophy than a hard nosed scientific theory.
I disagree about it being problematic to take introspective impressions as a foundation. If we were wanting to simply investigate the workings of the brain, then you're right, introspection is a pretty unreliable source. But if we're wanting to look at the nature of subjective experience itself, then we have to start with introspection so that we thoroughly know what it is we're trying to explain. Even if it's misleading or illusory in one way or another, we need to understand that manifest image (please correct me if I'm using that wrong) in order to be able to explain why it appears that way.
That's interesting that it's taken an idealist turn. What does that look like? I've not really got my head around what idealism is supposed to mean (besides Berkeley's version).
I think it's on the border between science and philosophy, but that's to be expected of consciousness studies imo, since it's pre paradigm. And in general the border between them is less solid than is often supposed.
I agree that the manifest image has to be accounted for. But IIT takes it as a guide to the substrate that a consciousness has to exist in. For all the reasons you discuss (like the arbitrary requirement about granularity) this just seems like assumptions to get a desired result.
Here's Birch's post about the idealism he found in recent IIT papers: https://philosophyofbrains.com/2023/09/12/consciousness-and-the-overton-window-of-science-part-ii.aspx (Note: the whole series of posts is worth checking out.)
Definitely every scientific theory of consciousness is both a philosophy and an empirical model. But as Birch covers in his series, IIT's philosophy is well outside of the current Overton window. Should that window shift? I'd say yes, if there are good reasons for it. I'm not sure what those would be in this case.
Oh wow, that post is quite shocking.
"The message is repeated over and over: my neurons do not truly exist. They exist in a derived, extrinsic sense—as stable appearances that other conscious observers, such as neurosurgeons, can use as handles of control over my experiences—but they do not intrinsically exist. What intrinsically exists is my conscious point of view."
This reminds me of a bit in Aristotle's Metaphysics where he says that the parts of a substance (eg parts of a living body) do not have substantial existence themselves, but only potential existence, so that there can be no substance composed of substances (for my own future reference, it's in Zeta 16). I don't think Aristotle was saying anything as radical as that paper was (pretty sure he wasn't an idealist), but it's interesting to see the similarity.
I suppose if IIT involves all matter being part of a conscious whole, then the similarity on this point is very close. And I think Aristotle kind of had a point: If we're looking at things in terms of their behaviour as part of a whole, it won't do to ignore the full context of the whole whole. You can't really understand a heart apart from its context within an animal for example.
The difficulty is, how do you draw the line for where a whole ends? I suppose that's where the exclusion principle comes in again, and Aristotle has his own criteria.
But if we take a relational metaphysics instead of a substantial one, then things look quite different. We can look at things in their full contexts without needing to draw any line on where their substance/whole ends, while still recognising that certain regions are more relevant than others.
This kind of relates to what you said in your recent post about what it's like to be a bat, seeing consciousness as kind of being "hooked up" to perceive yourself. Rather than seeing self-causal loops/consciousness as fundamentally different from our perception of external reality, as IIT does, we can see our self-perception and perception of the world as the same kind of thing. Hope that makes sense?
I think the Overton window can shift in response to empirical validation. Of course, metaphysical aspects of IIT and other theories of consciousness cannot be empirically tested directly, but for those aspects that can, these should imo count as evidence for the metaphysical elements. At least if one metaphysical system fits more nicely with all the evidence. As an example, imo evolution has major implications for metaphysics and the existence or nature of God.
I definitely think our reflective perceptions aren't fundamentally different than our external ones, so I'm onboard. When I'm trying to steelman IIT, I actually take all the talk of "having causal power over itself" as basically an alternate way of describing self reflection. But I find the language so convoluted and ambiguous, I can't be sure that's what they mean.